On non-monotonicity height of piecewise monotone functions (Q2029451)

The authors give a sufficient condition for piecewise monotone functions to have infinite height by introducing the concept of spanning interval. They also provide an algorithm to determine the spanning intervals for piecewise monotone (PM) functions. Using this algorithm, they compute maximal spanning interval for the following example. Let \(F_{\lambda}: [0, 1]\rightarrow [0, 1]\). It is given as \[ F_{\lambda}(x) = \begin{cases} 2\lambda x & \quad x \in [0, 1/2] ,\\ -2\lambda x+2\lambda &\quad x\in (1/2,3/4] , \\ 2\lambda x-\lambda & \quad x\in (3/4, 1], \end{cases} \] where \(\lambda \in (1/2, 1]\). The authors also provide another condition concerning periodic points. Let \(F\in \mathrm{PM}(I)\). Then, the non-monotonicity height \(H(F)\) is infinite if the function \(F\) satisfies either \(P_{4}(F)\neq \emptyset\) or \((P^{-}_{12}, P^{+}_{12}\cap S(F))\neq \emptyset\), where \(P^{-}_{12}:=\;\text{min}\{P_{1}, P_{2}\}\), \(P^{+}_{12}:=\text{max}\{P_{1}, P_{2}\}\) and \(P_{k}(F):=\{x\in I: x \;\text{is a periodic point of}\; F \; \text{of period}\; k, k\in \mathbb{N}\}\). Moreover, the authors find the density of all piecewise monotone functions with infinite and finite height, respectively. They prove that \(\mathrm{PM}_{\infty}(I)\) is dense in \(\mathrm{PM}(I)\), where \(\mathrm{PM}_{\infty}(I):=\{F\in \mathrm{PM}(I): H(F)=\infty\}\). Further, they provide the following example for which the set \(\mathrm{PM}_{<\infty}(I):=\{F\in \mathrm{PM}(I): H(F)<\infty\}\) is not dense in \(\mathrm{PM}(I)\). Let \(F: [0, 1]\rightarrow [0, 1]\) be defined by \[ F(x)= \begin{cases} 4x & \quad x \in [0, 1/4] \\ -4x+2 & \quad x\in (1/4,1/2] \\ 4x-2 & \quad x\in (1/2, 3/4] \\ -4x+4 & \quad x\in (3/4, 1]. \end{cases} \] In this case, the authors explore the variance of the height under composition, especially for functions of height 1 and infinity. Let \(F, G\in \mathrm{PM}(I)\) and \(H(F)=H(G)=1\). Then: \begin{itemize} \item[(1)] The following holds: (i) If \(S(F) \cap G(\text{int}K(F))=\emptyset\) and \(S(G)\cap \text{int}K(F) =\emptyset\), then \(H(F\circ G)=1\); (ii) If \(G(K(F))=K(G)\), \(F(K(G))=K(F)\) and \(S(G)\cap \text{int}K(F) \neq \emptyset\), then \(H(F\circ G)=\infty\). \item[(2)] If \(H(F\circ G)=1\), then \(\{\text{int}F\circ G(I)\cup G(\text{int}F\circ G(I))\}\cap \{S(F)\cup S(G)=\emptyset\}\). \end{itemize}